Polycomb chromobox (CBX) proteins participate in the polycomb repressive complex (PRC1) that mediates epigenetic gene silencing and endows PRC1 with distinct oncogenic or tumor suppressor functions in a cell-type–dependent manner. In this study, we report that inhibition of cell migration, invasion, and metastasis in colorectal carcinoma requires CBX4-mediated repression of Runx2, a key transcription factor that promotes colorectal carcinoma metastasis. CBX4 inversely correlated with Runx2 expression in colorectal carcinoma tissues, and the combination of high CBX4 expression and low Runx2 expression significantly correlated with overall survival, more so than either CBX4 or Runx2 expression alone. Mechanistically, CBX4 maintained recruited histone deacetylase 3 (HDAC3) to the Runx2 promoter, which maintained a deacetylated histone H3K27 state to suppress Runx2 expression. This function of CBX4 was dependent on its interaction with HDAC3, but not on its SUMO E3 ligase, its chromodomain, or the PRC1 complex. Disrupting the CBX4–HDAC3 interaction abolished Runx2 inhibition as well as the inhibition of cell migration and invasion. Collectively, our data show that CBX4 may act as a tumor suppressor in colorectal carcinoma, and strategies that stabilize the interaction of CBX4 with HDAC3 may benefit the colorectal carcinoma patients with metastases. Cancer Res; 76(24); 7277–89. ©2016 AACR.

Polycomb group (PcG) proteins have been documented to be major transcriptional repressors that mediate epigenetic gene silencing (1, 2). In general, PcG proteins organize to form two dominant polycomb -repressive complexes (PRC), PRC1 and PRC2 (3, 4), which are mainly involved in regulating development (5), senescence (6), stemness (7), and cancer progression (1, 4). PcG proteins are well known to be frequently dysregulated in various cancer types (8). However, because PcG complexes contain multiple subunits and PcG proteins carry out various functions, the exact regulatory elements of PcG proteins that regulate certain types of cancer have not yet been identified (4). Notably, several polycomb chromobox (CBX) proteins, including CBX2, CBX4, CBX6, CBX7, and CBX8, have also been shown to participate in the PRC1 complex and endow PRC1 with distinguish functions (9), suggesting that CBX proteins may act as an oncogene or tumor suppressor in a cell-type–dependent manner. For instance, CBX8 has been recently reported to exert paradoxical effects, promoting proliferation while suppressing metastasis, in colorectal carcinoma progression (10); CBX7 functions as a tumor suppressor in lung carcinoma by recruiting HDAC2 to the CCNE1 promoter to suppress CCNE1 expression (11), whereas it is an oncogene in gastric cancer and lymphoma (12, 13). Therefore, the function of each CBX protein in any cancer must be assessed separately.

CBX4 (also known as polycomb 2, Pc2) is a special chromobox protein because it is not only a transcriptional repressor but also a SUMO E3 ligase (14, 15). The N-terminal chromodomain and two SUMO-interacting motifs (SIM) confer CBX4 with polycomb- and SUMO E3 ligase–dependent functions, respectively (6, 16). For instance, CBX4 protects slow-cycling human epidermal stem cells (epSC) from senescence via its chromodomain, whereas it prevents active proliferation and terminal differentiation in epSCs via its SUMOylation activity (6). Recently, CBX4 has been identified as a new therapeutic target for hepatocellular carcinoma (HCC; refs. 9, 17), which is one of the most prevalent malignancies resistant to current chemotherapies or radiotherapies (18). CBX4 increases the transcriptional activity of HIF-1a, hypoxia-induced VEGF expression and angiogenesis by promoting HIF-1a SUMOylation at K391 and K477, and this function depends on the two SUMO-interacting motifs of CBX4 (9). In addition, high CBX4 expression predicts poor overall survival in patients with HCC (9, 17). Colorectal carcinoma is one of the most common causes of cancer-related death worldwide (19, 20). Thus, better understanding the molecular mechanisms of colorectal carcinoma tumorigenesis and identifying new therapeutic targets for colorectal carcinoma are urgent research objectives. Here, we provide evidence to show that CBX4 may serve as a tumor suppressor in colorectal carcinoma by recruiting HDAC3 to the Runx2 promoter to impede Runx2 expression and that this novel function of CBX4 is independent of its SUMO E3 ligase, its chromodomain, and the PRC1 complex.

Microarray data analysis

Briefly, samples (HCT116 cells transfected with control or CBX4 shRNA) were used to synthesize cRNA, and cRNA was labeled and hybridized to Gene Expression Hybridization Kit (cat# 5188-5242, Agilent Technologies) in Hybridization Oven (cat# G2545A, Agilent Technologies), according to the manufacturer's instructions, and samples were scanned by Agilent Microarray Scanner (cat# G2565CA, Agilent Technologies) with default settings (dye channel: green, scan resolution = 5 μm, PMT 100%, 10%, 16 bit). Data were extracted with Feature Extraction software 10.7 (Agilent Technologies). Raw data were normalized by Quantile algorithm, Gene Spring Software 11.0 (Agilent Technologies). Data are available via Gene Expression Omnibus (GEO) GSE87778, and in Supplementary Data S1.

Cell culture

The HCT116, DLD1, LOVO, SW620, SW480, and 293T embryonic kidney cells were obtained from the ATCC in 2013. The THC8307 cell line was a gift from the Institute of Hematology (CAMS & PUMC, Tianjin, China) in 2013. All cell lines were cultured in DMEM (Life Technologies) supplemented with 10% FBS (Life Technologies) with 5% CO2 at 37°C. All cell lines used in this study were authenticated using short-tandem repeat profiling less than 6 months ago when this project was initiated in 2013, and the cells have not been in culture for more than 2 months.

Plasmid construction

The flag-HDAC constructs were gifts from Prof. Binhua P. Zhou (University of Kentucky, Lexington, KY). pBABE-puro-CDM, pBABE-puro-ΔSIM1/2, and pBABE-puro-CBX4 were generously provided by Prof. Guo-Qiang Chen (Shanghai Jiao Tong University School of Medicine, China). Myc-tagged CDM, ΔSIM1/2, and CBX4 were cloned into the pCDNA3.1 vector. HDAC1, HDAC2, and HDAC3 were cloned into the pSIN-EF2-puro vector. The promoter regions of Runx1, Runx2, TGFB2, Slug, VIM, FN1, SOX2, VCAN, and ITGB8 were cloned into the pGL3-basic vector. The PLKO.1-puro vector was used to clone the shRNAs targeting CBX4, Runx2, or Slug. The sequences used for cloning the indicated shRNAs are shown in Supplementary Materials and Methods.

RNAi treatment

The sequence targeting HDAC3, 5′-GCATTGATGACCAGAGTTA-3′, has been used previously (21). Transfection was performed according to the manufacturer's instructions using Lipofectamine RNAiMAX transfection reagent (Invitrogen) and 50 nmol/L siRNA.

RNA extraction and qRT-PCR

Briefly, total RNA was isolated using TRIzol reagent (Invitrogen) according to the manufacturer's instructions. First-strand cDNA was synthesized using the Revert AidTM First Strand cDNA Synthesis Kit (MBI Fermentas). The primers used to amplify the indicated genes are shown in Supplementary Materials and Methods.

The luciferase reporter assay

Briefly, the cells were plated in 12-well plates at a density of 1.4 × 105 cells per well then transfected with 0.8 μg of promoter-luciferase plasmid. To normalize the transfection efficiency, the cells were also cotransfected with 8 ng of pRL-CMV (Renilla luciferase). After transfection for 48 hours, the luciferase activity was measured using a Dual-Luciferase Assay kit (Promega). Three independent experiments were performed, and the calculated means and SDs are presented. The primers used for cloning the indicated promoters are shown in Supplementary Materials and Methods.

Western blotting and coimmunoprecipitation

Briefly, cells were collected and lysed by RIPA buffer (150 mmol/L NaCl, 0.5% EDTA, 50 mmol/L Tris, 0.5% NP40) and centrifuged for 20 minutes at 12,000 rpm and 4°C. Fifty micrograms of harvested total protein was loaded and separated on the 8% SDS–polyacrylamide gradient gel. The proteins were then transferred onto polyvinylidene difluoride membranes and blocked with 5% non-fat milk for 2 hours at room temperature. The membranes were incubated with primary antibody and horseradish peroxidase–conjugated secondary antibody, and proteins were then detected using the ECL chemiluminescence system (Pierce).

For coimmunoprecipitation, the clarified supernatants were first incubated with anti-FLAG-agarose (Sigma Aldrich) gels for 2 hours or overnight at 4°C, and the precipitates were washed five times with RIPA. To investigate the interaction between endogenous CBX4 and HDAC3, the clarified supernatants were first incubated with an anti-CBX4 antibody for 2 hours at 4°C. Protein A/G-agarose was then added for 2 hours or overnight, and the precipitates were washed five times with RIPA and analyzed by Western blotting. The antibodies used in this work are shown in Supplementary Materials and Methods.

Chromatin immunoprecipitation assay

This procedure was performed as described by the ChIP kit (Millipore, 17-10085 & 17-10086). Briefly, 15-cm plates were seeded with cells of each of the tested cell lines and allowed to grow to 70%–80% confluence. To fix cells, complete cell fixative solution (1/10th the volume of growth medium volume) was added to the existing culture medium. The fixation reaction was stopped by adding stop solution (1/20th the volume of growth medium volume) to the existing culture medium. The cells were collected by centrifugation, and the nuclear pellet was resuspended in chromatin immunoprecipitation (ChIP) buffer. The cell lysate was subjected to sonication and then incubated with 5 μg of antibodies overnight, followed by incubation with the protein A/G agarose beads overnight at 4°C. Bound DNA–protein complexes were eluted, and cross-links were reversed after a series of washes. The purified DNA was resuspended in TE buffer for PCR. The primers for the indicated promoters are shown in Supplementary Materials and Methods.

MTT assay

A 3-(4, 5-Dimethylthiazol-2-yl)-2, 5-diphenyltetrazolium bromide (MTT) assay was used to measure cell viability. Briefly, HCT116, THC8307, or DLD1 cells were seeded at a density of 2,000 cells per well in a 96-well microplate. The cells were incubated with MTT for 4 hours, and the optical density (OD) was detected at 490 nm with the microplate reader once per day for 5 days. The results are presented as the mean ± SD of three independent experiments.

Soft agar colony formation assay

HCT116, THC8307, or DLD1 cells in a single-cell suspension were plated in 0.3% agarose over the 0.6% agarose bottom layer in 6-well plates at the density of 1 × 104 cells per well, and were incubated for 10 days. The colonies containing more than 50 cells were counted.

Migration and invasion assays

Transwell assays using Boyden chambers containing 24-well Transwell plates (BD Biosciences) with 8-mm pore size were used to evaluate the migration and invasiveness of cells. All experiments were performed in duplicate and repeated three times. For the migration assay, the cell culture inserts were seeded with 1.5 × 105 (HCT116 cells), 3 × 104 (THC8307 cells), and 1 × 105 (DLD1 cells) in 200 μL of serum-free DMEM without an extracellular matrix coating. DMEM medium containing 20% FBS was added to the bottom chamber. After approximately 24 hours of incubation, the cells on the bottom surface of the filter were fixed, stained, and examined using a microscope. For the invasion assay, the membrane was coated with 50 μL of 1:8 diluted Matrigel (BD Biosciences). After the Matrigel had solidified at 37°C for 2 hours, 3 × 105 (HCT116 cells), 5 × 104 (THC8307 cells), of 2 × 105 (DLD1 cells) in 200 μL of serum-free DMEM were added to the cell culture inserts, whereas the bottom chamber was filled with DMEM containing 20% FBS. The Boyden chamber was then incubated at 37°C in 5% CO2 for approximately 24 hours. The cells were then stained and observed as described for the migration assays.

Animal experiments

Animal care and experiments were performed in strict accordance with the “Guide for the Care and Use of Laboratory Animals” and the “Principles for the Utilization and Care of Vertebrate Animals” and were approved by the Animal Research Committee of Sun Yat-sen University Cancer Center.

All animals were obtained from the Shanghai Institutes for Biological Sciences (Shanghai, China). Lung and liver metastasis models were used. For the lung metastatic model, the indicated cells were harvested and washed twice with 1× PBS. The cells were then suspended in PBS. Approximately 2 × 106 cells in 150-μL PBS were injected into the tail veins of 5-week-old male athymic mice. For the liver metastatic model, the spleen was accessed via a 1-cm incision in the upper left lateral abdomen, and 2 × 106 cells suspended in 50 μL of PBS were injected into the distal tip of the spleen using an insulin syringe. The spleen was then placed back into the abdomen, and the abdominal cavity was closed with sutures. All mice were sacrificed 10 weeks after the injection, and the lungs or livers were harvested. The metastatic nodules in each lung or liver were counted.

Human tissue specimens

A total of 212 paraffin-embedded primary specimens were obtained from the recruited colorectal carcinoma patients. The patients were diagnosed according to their clinicopathologic characteristics at the Sun Yat-sen University Cancer Center (Guangzhou, China), from 2001 to 2009 (115 colon carcinoma patients with ages from 19 to 83 years, 97 rectum carcinoma patients with ages from 25 to 75 years). No patients had received radiotherapy and/or chemotherapy prior to surgery. Tumors were staged according to the Union for International Cancer Control TNM staging system. Resected specimens were macroscopically examined to determine the location and size of a tumor, and specimens for histology were fixed in 10% (v/v) formalin and processed for paraffin embedding. Informed consent was obtained from all patients and approved by the research medical ethics committee of Sun Yat-sen University Cancer Center.

Immunohistochemical staining

Immunohistochemical staining was performed on 3-μm sections. The primary antibodies against CBX4 or Runx2 were diluted 1:200 or 1:1,000, respectively, and then incubated at 4°C overnight in a humidified container. After three washes with PBS, the tissue slides were treated with a non-biotin horseradish peroxidase detection system according to manufacturer's instructions (Dako). The immunohistochemical staining was evaluated by two independent pathologists. The protein expression levels of CBX4 and Runx2 were evaluated on the basis of thirteen scores. Generally, the CBX4 and Runx2 signals were detected in the nucleus. The staining intensity of CBX4 or Runx2 was stratified into four classes in colorectal carcinoma tissues, namely 0, 1, 2, and 3, which were designated as absent, weak, moderate, and strong signals, respectively. The percentage of stained cells was categorized as 0, 1, 2, 3, and 4 to indicate no staining, 1%–10%, 11%–50%, 51%–80% and 81%–100% stained cells, respectively. The score for each tissue was calculated by multiplying the staining value by the percentage category value, and the average score from the two pathologists (M.F. Zhang and M. Li) was used as the final score. The immunohistochemical cut-off for high or low expression of indicated molecule was determined through the ROC curve analysis. The sensitivity and specificity for discriminating dead or alive was plotted as IHC score, thus generating a ROC curve. The cut-off value was established to be the point on the ROC curve where the sum of sensitivity and specificity was maximized. Cancers with scores above the obtained cut-off value were considered to have high expression of indicated molecule and vice versa.

Statistical analysis

The SPSS software (version 16.0, SPSS Inc.) was used for the statistical analysis. The significance of differences was assessed using two-tailed Student t test or a χ2 test, as appropriate. The relationship between CBX4 expression, Runx2 expression, and the clinicopathologic parameters was examined using the Pearson χ2 tests. The correlations between CBX4 expression, Runx2 expression, and overall survival curves were assessed using Kaplan–Meier plots and compared with the log-rank test. Univariate and multivariate Cox regression analyses were used to evaluate survival data. Differences were considered significant when the P values were <0.05.

Study approval

The animal experiments were approved by the Animal Research Committee of Sun Yat-sen University Cancer Center and were performed in accordance with the established guidelines. The use of human colorectal carcinoma tissues was reviewed and approved by the ethical committee of Sun Yat-sen University Cancer Center, and informed consent was obtained. The samples were retrospectively acquired from the surgical pathology archives of Sun Yat-sen University Cancer Center.

CBX4 negatively regulates cancer metastasis in colorectal carcinoma

CBX proteins may act as an oncogene or tumor suppressor in a cell-type–dependent manner (4), and we have recently reported that CBX8 exerts paradoxical effects in colorectal carcinoma, promoting proliferation while suppressing metastasis (10). Therefore, we sought to determine the role of each CBX protein in colorectal carcinoma metastasis because the molecular mechanisms of colorectal carcinoma progression urgently need to be elucidated (19, 20). Initially, following the strategy used in HCC (9), we ectopically expressed Flag-tagged CBX2, CBX4, CBX6, CBX7, or CBX8 in DLD1 cells and found that only CBX4 suppressed cell migration (Supplementary Fig. S1A–S1C). In fact, the CBX4 protein levels differed among the tested colorectal carcinoma cell lines (Fig. 1A). Next, we used two pairs of short hairpin RNAs (shRNA) to dramatically reduce endogenous CBX4 expression (Supplementary Fig. S2A and S2B). This knockdown significantly enhanced the migration and invasion of both HCT116 and THC8307 cells, which endogenously express high levels of CBX4 (Fig. 1B and C; Supplementary Fig. S2C and S2D). Conversely, the knockdown of CBX4 marginally affected both cell viability and soft agar colony formation ability in these cell lines (Supplementary Fig. S2E–S2H). In contrast, the stably ectopic expression of CBX4 dramatically inhibited the migration and invasion in DLD1 cells, which endogenously express low levels of CBX4 (Fig. 1D; Supplementary Fig. S3A and S3B). Moreover, the stable ectopic expression of CBX4 did affect neither the viability nor soft agar colony formation ability of these cells (Supplementary Fig. S3C and S3D). Knocking down CBX4 not only consistently increased the number of metastatic nodules but also enhanced the tumor size (Fig. 1E and F) when HCT116 cells were injected into the tail veins of mice, and this increase was recapitulated in the hepatic metastasis in vivo animal model using these cells (Supplementary Fig. S4A and S4B). Taken together, CBX4 negatively regulates cell migration, invasion, and cancer metastasis, but not the cell proliferation, in colorectal carcinoma.

Figure 1.

CBX4 represses the migration, invasion, and metastasis of colorectal carcinoma cells. A, The CBX4 protein level was visualized in the indicated cell lines by a Western blot analysis. B–D, The migration and invasion of HCT116 (B), THC8307 (C), and DLD1 (D) cells were determined as described in Materials and Methods. Bars, SD. The results are expressed as the mean ± SD of three independent experiments. *, P < 0.05; **, P < 0.01 using Student t test. E and F, The lung metastasis of the indicated stable cell lines was measured in vivo in nude mice as described in Materials and Methods. E, Representative results of gross (left) and hematoxylin and eosin (H&E) staining (middle, scale, ×40; right, scale, ×200) of the metastatic lung nodules. F, The statistical results (n = 8). Bars, SD. *, P < 0.05 using Student t test.

Figure 1.

CBX4 represses the migration, invasion, and metastasis of colorectal carcinoma cells. A, The CBX4 protein level was visualized in the indicated cell lines by a Western blot analysis. B–D, The migration and invasion of HCT116 (B), THC8307 (C), and DLD1 (D) cells were determined as described in Materials and Methods. Bars, SD. The results are expressed as the mean ± SD of three independent experiments. *, P < 0.05; **, P < 0.01 using Student t test. E and F, The lung metastasis of the indicated stable cell lines was measured in vivo in nude mice as described in Materials and Methods. E, Representative results of gross (left) and hematoxylin and eosin (H&E) staining (middle, scale, ×40; right, scale, ×200) of the metastatic lung nodules. F, The statistical results (n = 8). Bars, SD. *, P < 0.05 using Student t test.

Close modal

CBX4 occupies the Runx2 promoter and represses its expression

As a transcription repressor, CBX4 frequently targets the promoters of downstream genes and mediates their epigenetic modification (6). To investigate the molecular mechanism of CBX4 in colorectal carcinoma, the global transcriptomes were analyzed in HCT116 cells in which CBX4 was knocked down with shRNA, and these transcriptomes were compared with those of HCT116 cells transfected with scrambled shRNA (GEO accession number: GSE87778). Intriguingly, gene ontology (GO) or Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analyses indicated that the genes upregulated by knocking down CBX4 were mainly associated with cell adhesion, migration, and the extracellular matrix (Fig. 2A; Supplementary Fig. S5A and S5B; Supplementary Data S1). Indeed, the expression of factors that are well known to be related to metastasis, such as Slug, vimentin, and fibronectin (FN), increased by knocking down CBX4 in colorectal carcinoma cells, as shown in both the qRT-PCR assay (Fig. 2B) and the Western blots (Fig. 2C). Accordingly, the ectopic expression of CBX4 decreased the levels of Slug, vimentin, and FN in DLD1 cells (Fig. 2D).

Figure 2.

CBX4 directly downregulates Runx2 by occupying its promoter. A, Representative heatmaps from a global comparative transcriptome analysis indicating genes that are upregulated upon CBX4 depletion. B, The relative mRNA levels of the indicated genes were normalized to the GAPDH level in the HCT116 cells stably transfected with control or shRNA-CBX4 as determined by qRT-PCR. Bars, SD. The results are expressed as the mean ± SD of three independent experiments. *, P < 0.05; **, P < 0.01 using Student t test. C, The indicated proteins were analyzed by Western blotting in HCT116 and THC8307 cells stably expressing control or shRNA-CBX4, as indicated. D, The indicated proteins were analyzed by Western blotting in DLD1 cells stably expressing pBABE-Vector or pBABE-CBX4. E, The ChIP assay was performed with HCT116 cells using an anti-CBX4 antibody or IgG antibody, as indicated. The p16 and GAPDH promoters were used as the positive and negative controls, respectively. F, The indicated proteins were analyzed by Western blotting in HCT116 and THC8307 cells stably expressing control or shRNA-CBX4 and in DLD1 cells stably expressing pBABE-Vector or pBABE-CBX4.

Figure 2.

CBX4 directly downregulates Runx2 by occupying its promoter. A, Representative heatmaps from a global comparative transcriptome analysis indicating genes that are upregulated upon CBX4 depletion. B, The relative mRNA levels of the indicated genes were normalized to the GAPDH level in the HCT116 cells stably transfected with control or shRNA-CBX4 as determined by qRT-PCR. Bars, SD. The results are expressed as the mean ± SD of three independent experiments. *, P < 0.05; **, P < 0.01 using Student t test. C, The indicated proteins were analyzed by Western blotting in HCT116 and THC8307 cells stably expressing control or shRNA-CBX4, as indicated. D, The indicated proteins were analyzed by Western blotting in DLD1 cells stably expressing pBABE-Vector or pBABE-CBX4. E, The ChIP assay was performed with HCT116 cells using an anti-CBX4 antibody or IgG antibody, as indicated. The p16 and GAPDH promoters were used as the positive and negative controls, respectively. F, The indicated proteins were analyzed by Western blotting in HCT116 and THC8307 cells stably expressing control or shRNA-CBX4 and in DLD1 cells stably expressing pBABE-Vector or pBABE-CBX4.

Close modal

A luciferase reporter assay was then used to further identify the direct downstream target of CBX4. Among the upregulated genes shown in Fig. 2B, only the promoter activity of Runt-related transcription factor 2 (Runx2) was significantly increased upon CBX4 knockdown (Supplementary Fig. S5C). Furthermore, as shown in Fig. 2E, CBX4 was associated with the Runx2 promoter according to a chromatin immunoprecipitation (ChIP) assay using the p16 (INK4a/ARF) promoter and the GAPDH promoter as positive and negative controls (6, 10). As shown in Fig. 2F, changes in CBX4 expression (knockdown and ectopic expression), respectively, increased and decreased the Runx2 levels in colorectal carcinoma cells. Taken together, these data indicate that CBX4 can downregulate Runx2 by occupying its promoter in colorectal carcinoma cells.

The effect of CBX4 on colorectal carcinoma metastasis depends on Runx2

Next, we tested the dependence of the observed increase in metastasis due to CBX4 knockdown on Runx2. Knocking down Runx2 using two pairs of shRNAs reduced the expression of Slug (Supplementary Fig. S6A and S6B), a downstream gene of Runx2 (22), and, as expected, silencing Runx2 abolished the increase in Slug at both the protein and mRNA levels induced by knocking down CBX4 in colorectal carcinoma cells (Fig. 3A and B). More importantly, silencing Runx2 abrogated the increase in cell migration, invasion, and metastasis induced by knocking down CBX4 in HCT116 cells (Fig. 3C–E; Supplementary Fig. S7). Consistently, silencing Slug also reversed the increase of cell migration and invasion induced by knocking down CBX4 (Supplementary Fig. S8). These results demonstrate that knocking down CBX4 promotes metastasis by abrogating its inhibition of Runx2.

Figure 3.

The promotion of cell migration and invasion after CBX4 knockdown primarily depends on Runx2. A and B, HCT116 cells stably expressing control, shRNA-CBX4, shRNA-Runx2, or both, as indicated, were analyzed by Western blotting (A) and qRT-PCR (B). C, A cell migration and invasion assay was used in HCT116 cells stably expressing control, shRNA-CBX4, shRNA-Runx2, or both, as indicated. Bars, SD. The results are expressed as the mean ± SD of three independent experiments. *, P < 0.05; **, P < 0.01 using Student t test. D and E, An in vivo lung metastasis model was established in nude mice using HCT116 cells stably expressing control, shRNA-CBX4, shRNA-Runx2, or both, as indicated, as described in Materials and Methods. D, Representative results of gross (left) and H&E staining (middle, scale, ×40; right, scale, ×200) of metastatic lung nodules. E, The statistical results (n = 8). Bars, SD. *, P < 0.05; **, P < 0.01 using Student t test.

Figure 3.

The promotion of cell migration and invasion after CBX4 knockdown primarily depends on Runx2. A and B, HCT116 cells stably expressing control, shRNA-CBX4, shRNA-Runx2, or both, as indicated, were analyzed by Western blotting (A) and qRT-PCR (B). C, A cell migration and invasion assay was used in HCT116 cells stably expressing control, shRNA-CBX4, shRNA-Runx2, or both, as indicated. Bars, SD. The results are expressed as the mean ± SD of three independent experiments. *, P < 0.05; **, P < 0.01 using Student t test. D and E, An in vivo lung metastasis model was established in nude mice using HCT116 cells stably expressing control, shRNA-CBX4, shRNA-Runx2, or both, as indicated, as described in Materials and Methods. D, Representative results of gross (left) and H&E staining (middle, scale, ×40; right, scale, ×200) of metastatic lung nodules. E, The statistical results (n = 8). Bars, SD. *, P < 0.05; **, P < 0.01 using Student t test.

Close modal

CBX4 and Runx2 expression inversely correlates in colorectal carcinoma tissues, and CBX4 may be a new marker of colorectal carcinoma

The clinical significance of the regulation of Runx2 by CBX4 was then evaluated in 212 colorectal carcinoma tissues by immunohistochemical (IHC) staining. Using both anti-CBX4 and anti-Runx2 antibodies, which were highly specific (Supplementary Fig. S9A and S9B), 120 of 212 tissues were shown to express high levels of CBX4, whereas 104 of 212 tissues expressed high levels of Runx2. Moreover, the levels of these two proteins inversely correlated (Fig. 4A; Supplementary Table S1). In addition, higher levels of CBX4 and Runx2 were associated with better and poorer overall survival in colorectal carcinoma patients, respectively (Fig. 4B and C). More interestingly, the combination of a high level of CBX4 and a low level of Runx2 more accurately predicted the overall survival of colorectal carcinoma patients (Fig. 4D). In addition, the χ2 test revealed that the CBX4 level was related to differentiation (P = 0.010) and vital status (P = 0.019; Supplementary Table S2), and the univariate and multivariate analyses revealed that the CBX4 expression level (P = 0.047), with a HR of 0.592 and a 95% CI of 0.353–0.992, was an independent prognostic factor in patients with colorectal carcinoma (Supplementary Table S3). These results indicate that CBX4 may be a new marker to predict clinical outcomes in colorectal carcinoma patients.

Figure 4.

The combination of CBX4 with Runx2 correlates with clinical prognosis in colorectal carcinoma. A, Representative immunohistochemical staining of CBX4 and Runx2 from 212 paraffin-embedded colorectal carcinoma tissues. Scale bar, 100 μm. B and C, Overall survival curves were generated on the basis of the protein level of CBX4 (B) or Runx2 (C) in 212 paraffin-embedded colorectal carcinoma tissues. D, Comparison of the overall survival between 57 CBX4high/Runx2low and 73 CBX4low/Runx2high tissues. Actuarial probabilities were calculated using the Kaplan–Meier method and compared using the log-rank test.

Figure 4.

The combination of CBX4 with Runx2 correlates with clinical prognosis in colorectal carcinoma. A, Representative immunohistochemical staining of CBX4 and Runx2 from 212 paraffin-embedded colorectal carcinoma tissues. Scale bar, 100 μm. B and C, Overall survival curves were generated on the basis of the protein level of CBX4 (B) or Runx2 (C) in 212 paraffin-embedded colorectal carcinoma tissues. D, Comparison of the overall survival between 57 CBX4high/Runx2low and 73 CBX4low/Runx2high tissues. Actuarial probabilities were calculated using the Kaplan–Meier method and compared using the log-rank test.

Close modal

CBX4 may act in colorectal carcinoma by decreasing H3K27 acetylation (H3K27-Ac) at the Runx2 promoter

Because the functions of CBX proteins primarily depend on their chromodomain (23) and because SUMO E3 ligase activity is critical for the key roles of CBX4 in HCC (9, 24), we sought to investigate whether the functions of CBX4 in colorectal carcinoma depend on its SUMO E3 ligase or its polycomb role. As illustrated in Fig. 5A, the mutant of ΔSIM1/2-CBX4, in which the SUMO-interacting motifs (ΔSIM1/2) was depleted, lacked SUMO E3 ligase activity (9, 25), whereas the chromodomain mutant of CDM-CBX4, which harbors a CBX4F11A and CBX4W35L double mutation, no longer bound trimethylated H3K27 (H3K27-Me3) but maintained the proper structure of the protein (9, 16). Using DLD1 cells that stably overexpressed wild-type CBX4 (WT-CBX4), ΔSIM1/2-CBX4, or CDM-CBX4, we unexpectedly found that both mutants of CBX4 similarly inhibited Runx2 and its downstream effector Slug, and cell migration and invasion among these cell lines were comparable with the migration and invasion of WT-CBX4 cells (Fig. 5B–D; Supplementary Fig. S10). Consistent with these results, knocking down CBX4 had little effect on the H3K27-Me3 of the Runx2 promoter in HCT116 cells. However, the binding of H3K27-Ac, a super-enhancer marker, in the Runx2 promoter was significantly enhanced by knocking down CBX4 in these cells, which was confirmed by the dramatic increase of RNA polymerase II (Pol II) in the Runx2 promoter (Fig. 5E). Taken together, CBX4 may repress Runx2 expression, cell migration, and invasion in colorectal carcinoma by decreasing H3K27-Ac at the Runx2 promoter locus, and this effect is independent of both the SUMO E3 ligase activity and the chromodomain of CBX4.

Figure 5.

A decrease in the H3K27-Ac marker on the Runx2 promoter is responsible for the functions of CBX4 in colorectal carcinoma. A, The schematic illustration for CBX4 and its mutants used in this study. B and C, The indicated proteins or genes were analyzed by Western blotting (B) or qRT-PCR (C) in DLD1 cells stably overexpressing vector, WT-CBX4, ΔSIM1/2-CBX4, or CDM-CBX4, as indicated. Bars, SD. The results are expressed as the mean ± SD of three independent experiments. *, P < 0.05; **, P < 0.01; ***, P < 0.001 using Student t test. D, DLD1 cells stably overexpressing vector, WT-CBX4, ΔSIM1/2-CBX4, or CDM-CBX4, as indicated, were subjected to migration and invasion assays, as described in Materials and Methods. Bars, SD. The results are expressed as the mean ± SD of three independent experiments. **, P < 0.01; ***, P < 0.001 using Student t test. E, The ChIP-qPCR analysis of the occupancies of H3K27-Ac, H3K27-Me3, and Pol II on the Runx2 promoter in stably transfected HCT116 cells, as indicated. Bars, SD. The results are expressed as the mean ± SD of three independent experiments. *, P < 0.05 using Student t test.

Figure 5.

A decrease in the H3K27-Ac marker on the Runx2 promoter is responsible for the functions of CBX4 in colorectal carcinoma. A, The schematic illustration for CBX4 and its mutants used in this study. B and C, The indicated proteins or genes were analyzed by Western blotting (B) or qRT-PCR (C) in DLD1 cells stably overexpressing vector, WT-CBX4, ΔSIM1/2-CBX4, or CDM-CBX4, as indicated. Bars, SD. The results are expressed as the mean ± SD of three independent experiments. *, P < 0.05; **, P < 0.01; ***, P < 0.001 using Student t test. D, DLD1 cells stably overexpressing vector, WT-CBX4, ΔSIM1/2-CBX4, or CDM-CBX4, as indicated, were subjected to migration and invasion assays, as described in Materials and Methods. Bars, SD. The results are expressed as the mean ± SD of three independent experiments. **, P < 0.01; ***, P < 0.001 using Student t test. E, The ChIP-qPCR analysis of the occupancies of H3K27-Ac, H3K27-Me3, and Pol II on the Runx2 promoter in stably transfected HCT116 cells, as indicated. Bars, SD. The results are expressed as the mean ± SD of three independent experiments. *, P < 0.05 using Student t test.

Close modal

HDAC3 associates with CBX4 and specifically represses the Runx2 promoter via histone deacetylation

Histone deacetylase (HDAC) is mainly responsible for the deacetylation of histone H3K27 (26), and HDAC2 has been reported to be recruited to the CCNE1 promoter by CBX7 to suppress CCNE1 expression in lung carcinoma (11). We attempted to identify HDACs that may be recruited to the Runx2 promoter by CBX4 to repress Runx2 expression. Interestingly, among all 11 HDACs (from HDAC1 to HDAC11), CBX4 strongly bound HDAC1, HDAC2, HDAC3, and HDAC9 when each HDAC was tagged with Flag and cotransfected with Myc-tagged CBX4 into 293T cells (Supplementary Fig. S11). Furthermore, HDAC1, HDAC2, and HDAC3, but not HDAC9, were detectable in the tested colorectal carcinoma cell lines (Supplementary Fig. S12). Moreover, HDAC3, but not HDAC1 or HDAC2, dramatically reduced the Runx2 promoter activity (Fig. 6A). Indeed, as shown in Fig. 6B and C, Runx2 expression was significantly reduced at both the mRNA and protein levels in HCT116 cells stably overexpressing HDAC3, but not in HCT116 cells stably overexpressing HDAC1. Accordingly, the interaction between CBX4 and HDAC3 was already evident at endogenous expression levels (Fig. 6D), and both mutants of CBX4-ΔSIM1/2 and CBX4-CDM did not affect its binding with HDAC3 (Supplementary Fig. S13). Collectively, these data show that CBX4 may interact with HDAC3 to inhibit the Runx2 promoter activity and repress Runx2 expression.

Figure 6.

CBX4 recruits HDAC3 to suppress the expression of Runx2. A, HCT116 cells were cotransfected with the Runx2-Luc reporter and Flag-HDAC1, Flag-HDAC2, or Flag-HDAC3 for 48 hours and then subjected to a luciferase activity assay as described in Materials and Methods. B and C, HCT116 cells stably expressing pSIN-Vector, pSIN-HDAC1, or pSIN-HDAC3 were analyzed by qRT-PCR (B) and Western blotting (C). Bars, SD. The results are expressed as the mean ± SD of three independent experiments. ***, P < 0.001 using Student t test. D, The co-IP assay was performed in HCT116 cells expressing endogenous levels of CBX4 using anti-CBX4 antibody or anti-IgG antibody as indicated. E, HCT116 cells stably expressing control or shRNA-CBX4 were cotransfected with the Runx2-Luc reporter and Flag-HDAC3 for 48 hours and then subjected to a luciferase activity assay as described in Materials and Methods. Bars, SD. The results are expressed as the mean ± SD of three independent experiments. **, P < 0.01 using Student t test. F, The ChIP-qPCR analysis of the occupancies of HDAC1, HDAC2, or HDAC3 on the Runx2 promoter in HCT116 cells stably expressing control or shRNA-CBX4. Bars, SD. The results are expressed as the mean ± SD of three independent experiments. **, P < 0.01 using Student t test. G and H, DLD1 cells stably overexpressing Vector or CBX4 were transfected with control or HDAC3 siRNA for 48 hours, then the indicated proteins were analyzed by Western blotting (G) and the cell migration and invasion were determined as described in Materials and Methods (H).

Figure 6.

CBX4 recruits HDAC3 to suppress the expression of Runx2. A, HCT116 cells were cotransfected with the Runx2-Luc reporter and Flag-HDAC1, Flag-HDAC2, or Flag-HDAC3 for 48 hours and then subjected to a luciferase activity assay as described in Materials and Methods. B and C, HCT116 cells stably expressing pSIN-Vector, pSIN-HDAC1, or pSIN-HDAC3 were analyzed by qRT-PCR (B) and Western blotting (C). Bars, SD. The results are expressed as the mean ± SD of three independent experiments. ***, P < 0.001 using Student t test. D, The co-IP assay was performed in HCT116 cells expressing endogenous levels of CBX4 using anti-CBX4 antibody or anti-IgG antibody as indicated. E, HCT116 cells stably expressing control or shRNA-CBX4 were cotransfected with the Runx2-Luc reporter and Flag-HDAC3 for 48 hours and then subjected to a luciferase activity assay as described in Materials and Methods. Bars, SD. The results are expressed as the mean ± SD of three independent experiments. **, P < 0.01 using Student t test. F, The ChIP-qPCR analysis of the occupancies of HDAC1, HDAC2, or HDAC3 on the Runx2 promoter in HCT116 cells stably expressing control or shRNA-CBX4. Bars, SD. The results are expressed as the mean ± SD of three independent experiments. **, P < 0.01 using Student t test. G and H, DLD1 cells stably overexpressing Vector or CBX4 were transfected with control or HDAC3 siRNA for 48 hours, then the indicated proteins were analyzed by Western blotting (G) and the cell migration and invasion were determined as described in Materials and Methods (H).

Close modal

We also examined whether the HDAC3-mediated suppression of the Runx2 promoter activity depends on CBX4. As shown in Fig. 6E, the knockdown of CBX4 abolished the inhibition of Runx2 promoter activity by HDAC3. Moreover, using the ChIP-qPCR assay, both HDAC1 and HDAC3, but not HDAC2, bound the Runx2 promoter, and the binding affinity of HDAC3, but not HDAC1, with the Runx2 promoter was dramatically decreased and increased by knocking down and ectopic expression of CBX4, respectively (Fig. 6F; Supplementary Fig. S14A and S14B). Indeed, silencing HDAC3 could reverse the suppression of Runx2 and H3K27-Ac mediated by overexpression of CBX4 (Fig. 6G). However, cell migration and invasion were not further enhanced by the overexpression of CBX4 in DLD1 cells silenced of HDAC3, as the knockdown of HDAC3 already inhibited the cell migration and invasion in DLD1 cells (Fig. 6H; Supplementary Fig. S15). Therefore, HDAC3 is essential for the functions of CBX4 in colorectal carcinoma. These results demonstrate that CBX4 represses Runx2 expression by recruiting HDAC3 to sustain the deacetylation status of H3K27 at the Runx2 promoter.

The suppression of Runx2, cell migration, and invasion by CBX4 depends on its binding with HDAC3 in colorectal carcinoma

To further assess the importance of the physical interaction with HDAC3 for the functions of CBX4, we mapped the interaction regions of CBX4 for their ability to bind with HDAC3. A series of fragments of CBX4 were generated, as illustrated in Fig. 7A, and their interaction with HDAC3 was tested. We found three separate regions of CBX4, residues 161–179 (I), 260–269 (II), and 551–560 (III), that are important for its binding with HDAC3 (Fig. 7A; Supplementary Fig. S16A–S16D). To determine the key regions of CBX4 that are essential for this interaction, we further generated several mutants of CBX4 by differently depleting regions I, II, and III, as illustrated in Fig. 7B. The binding of HDAC3 with CBX4 was completely abolished by ΔTM-CBX4, which lacked these three regions, and this interaction was only partially diminished by ΔDM1-CBX4 and ΔDM2-CBX4, which each only lacked two of these three regions, whereas ΔDM3-CBX4 could not abolish this interaction, as shown in Fig. 7C. Notably, region III of CBX4 is also involved in transcriptional silencing and binding to other PcG proteins (27). Indeed, ΔC-CBX4 lacked region III and was able to abolish the binding of CBX4 with Bmi1 (Supplementary Fig. S17A), which is the indicator for CBX4 within the PRC1 complex. Conversely, ΔDM1-CBX4, which lacked both regions I and II, did not affect the binding of CBX4 with Bmi1 (Supplementary Fig. S17B). Because region III of CBX4 is required for CBX4 to participate in the PRC1 complex (14, 28), we investigated whether the participation of CBX4 in the PRC1 complex is crucial for the function of CBX4 in colorectal carcinoma. Using DLD1 cells stably overexpressing WT-CBX4, ΔDM1-CBX4, ΔC-CBX4, or ΔTM-CBX4, we found that both WT-CBX4 and ΔC-CBX4 similarly inhibited Runx2, cell migration, and invasion. In contrast, these effects were abrogated in both ΔDM1-CBX4 and ΔTM-CBX4 cells (Fig. 7D and E; Supplementary Fig. S18). Taken together, these results strongly indicate that the suppression of Runx2, cell migration, and invasion in colorectal carcinoma by CBX4 strongly depend on its binding with HDAC3, but not on its binding with the PRC1 complex.

Figure 7.

The key regions that facilitate binding between CBX4 and HDAC3 are required for the functions of CBX4 in colorectal carcinoma. A, Schematic illustration of CBX4 mutants with the indicated HDAC3-binding capability. +, binding; −, no binding based on the results shown in Supplementary Fig. S16A–S16D. The three HDAC3 binding regions of CBX4 are indicated by a blue box. I, residues 161–170; II, residues 260–269; and III, residues 551–560. B, The schematic illustration of CBX4 mutants in which the indicated regions were depleted; ΔC represents the depletion of III, ΔDM1 represents the depletion of I and II, DM2 represents the depletion of II and III, ΔDM3 indicates the depletion of I and III, and ΔTM represents the depletion of I, II, and III. C, HEK293T cells were cotransfected with Myc-HDAC3 with vector, Flag-CBX4-WT, Flag-CBX4-ΔDM1, Flag-CBX4-ΔTM, Flag-CBX4-ΔDM2, or Flag-CBX4-ΔDM3 for 48 hours and then lysed and analyzed by immunoprecipitation (IP) using Flag-agarose and Western blotting. D, The indicated proteins were analyzed by Western blotting in DLD1 cells stably expressing vector, CBX4-WT, CBX4-ΔDM1, CBX4-ΔTM, or CBX4–ΔC, as indicated. E, Cell migration and invasion were assessed in DLD1 cells stably expressing vector, CBX4-WT, CBX4-ΔDM1, CBX4-ΔTM, or CBX4–ΔC, as indicated and as described in Materials and Methods. Bars, SD. The results are expressed as the mean ± SD of three independent experiments. *, P < 0.05; **, P < 0.01; ***, P < 0.001, using Student t test. F, A proposed model illustrating the regulation of colorectal carcinoma metastasis by CBX4. The repressive complex containing CBX4 and HDAC3 sustains the deacetylation status of H3K27 in the Runx2 promoter to repress Runx2 expression in colorectal carcinoma. However, the downregulation of CBX4 in colorectal carcinoma reduces HDAC3 in this repressive complex to maintain a high level of H3K27-Ac, which consequently activates Runx2 transcription to promote metastasis. CRC, colorectal carcinoma.

Figure 7.

The key regions that facilitate binding between CBX4 and HDAC3 are required for the functions of CBX4 in colorectal carcinoma. A, Schematic illustration of CBX4 mutants with the indicated HDAC3-binding capability. +, binding; −, no binding based on the results shown in Supplementary Fig. S16A–S16D. The three HDAC3 binding regions of CBX4 are indicated by a blue box. I, residues 161–170; II, residues 260–269; and III, residues 551–560. B, The schematic illustration of CBX4 mutants in which the indicated regions were depleted; ΔC represents the depletion of III, ΔDM1 represents the depletion of I and II, DM2 represents the depletion of II and III, ΔDM3 indicates the depletion of I and III, and ΔTM represents the depletion of I, II, and III. C, HEK293T cells were cotransfected with Myc-HDAC3 with vector, Flag-CBX4-WT, Flag-CBX4-ΔDM1, Flag-CBX4-ΔTM, Flag-CBX4-ΔDM2, or Flag-CBX4-ΔDM3 for 48 hours and then lysed and analyzed by immunoprecipitation (IP) using Flag-agarose and Western blotting. D, The indicated proteins were analyzed by Western blotting in DLD1 cells stably expressing vector, CBX4-WT, CBX4-ΔDM1, CBX4-ΔTM, or CBX4–ΔC, as indicated. E, Cell migration and invasion were assessed in DLD1 cells stably expressing vector, CBX4-WT, CBX4-ΔDM1, CBX4-ΔTM, or CBX4–ΔC, as indicated and as described in Materials and Methods. Bars, SD. The results are expressed as the mean ± SD of three independent experiments. *, P < 0.05; **, P < 0.01; ***, P < 0.001, using Student t test. F, A proposed model illustrating the regulation of colorectal carcinoma metastasis by CBX4. The repressive complex containing CBX4 and HDAC3 sustains the deacetylation status of H3K27 in the Runx2 promoter to repress Runx2 expression in colorectal carcinoma. However, the downregulation of CBX4 in colorectal carcinoma reduces HDAC3 in this repressive complex to maintain a high level of H3K27-Ac, which consequently activates Runx2 transcription to promote metastasis. CRC, colorectal carcinoma.

Close modal

In this report, we demonstrated that CBX4 may function as a tumor suppressor to impair Runx2 transcription by recruiting HDAC3 to deacetylate histone H3K27-Ac at the Runx2 promoter in colorectal carcinoma. This evidence is the first to show that CBX4 exerts a noncanonical function by recruiting HDAC3 to mediate histone deacetylation, which inhibits gene transcription. Furthermore, this novel function of CBX4 is independent on its SUMO E3 ligase, its chromodomain, and the PRC1 complex.

Traditionally, the PRC2 complex promotes H3K27-Me3, and the PRC1 complex then binds to H3K27-Me3 via its chromobox (CBX) proteins (1, 4). Therefore, CBX4 participates in the PRC1 complex, and its functions largely depend on the chromodomain (4, 6). For instance, CBX4 has been shown to repress transcription by binding the promoters of several target genes, such as p16 (INK4a/ARF; ref. 6) and GATA4/GATA6 (29). However, it has been reported that certain CBX proteins, such as CBX4 (6, 9) and CBX8 (30), can associate with protein complex other than PRC1, thereby playing a PRC1-independent role in transcriptional regulation (23, 30). Intriguingly, we have shown here that CBX4 represses Runx2 transcription in colorectal carcinoma by inhibiting H3K27-Ac markers but not promoting H3K27-Me3 markers and by recruiting HDAC3 at the Runx2 promoter, which consequently impairs cell migration and invasion. Both the recruitment of HDAC3 and the repression of cell migration and invasion by CBX4 are independent of its SIMs, its chromodomain, and the PRC1 complex. These new findings strongly suggest that the interaction with HDAC3 is pivotal for the function of CBX4 in colorectal carcinoma metastasis, as proposed in Fig. 7F. Notably, ΔC-CBX4 had a much higher protein level than WT-CBX4, indicating that the C-terminal 10 amino acids may be critical for CBX4 protein stability. Because these 10 amino acids are essential for binding with the PRC1 core subunits, such as Ring1a, Ring1b and Bmi1, we speculate that the E3 ligase activity of Ring1a and/or Ring1b or an unknown E3 ligase may degrade CBX4 in the intact PRC1 complex. This phenomenon is currently under investigation in our laboratory.

Because of its exclusive SUMO E3 ligase activity, CBX4 also promotes the SUMOylation of HIF-1a, CtBP1, and ZEB1 (9, 15, 31). Recently, CBX4 has been proposed to play an oncogenic role in HCC by positively regulating proliferation (17), angiogenesis (9) and cancer metastasis (24), and this function appears to be involved in two SIMs of CBX4 (9). However, our results showed that CBX4 may be a tumor suppressor in colorectal carcinoma by targeting Runx2 because knocking down and ectopically overexpressing CBX4 in colorectal carcinoma cell lines, respectively, increased and decreased cell migration, invasion, and metastasis. The reverse correlation between CBX4 and Runx2 was observed in colorectal carcinoma tissues, and the combination of CBX4high and Runx2low is a good marker to better predict the overall survival of patients with colorectal carcinoma. In fact, CBX4 exerts an antioncogenic effect by repressing the expression of c-myc and cellular transformation (32). Furthermore, CBX4 associates with E2F and Rb to block entry into mitosis by inhibiting cyclin A and cdc2 transcription, which in turn negatively regulate cell proliferation (33). Therefore, CBX4 can likely exert both oncogenic and antioncogenic functions depending on the cancer type and its interacting partners. This pleiotropic nature is not surprising because CBX7 also plays diverse roles in different cancer types (11–13, 34). CBX7 recruits HDAC2 to the promoter of the CCNE1 gene and inhibits histone acetylation to consequently suppress CCNE1 expression and lung carcinoma progression (11). Conversely, CBX7 is oncogenic in lymphoma and gastric cancer because it represses the p16 (INK4a/ARF) tumor suppressor locus (12, 13).

Cancer metastasis is a common cause of cancer-related death, and agents targeting the epithelial-to-mesenchymal transition (EMT) or angiogenesis, which are crucial for metastasis, are potential drugs for cancer therapy (22, 35). Runx family proteins play an important role in cancer progression (35). Most Runx family genes, such as Runx1 and Runx3, have been shown to have tumor suppressor activities (35–37), whereas Runx2 is unique due to its oncogenic features. Runx2 has been well documented to induce invasion and metastasis in different cancers, such as breast cancer (35), prostate cancer (38), thyroid cancer (39, 40), and colorectal carcinoma (19, 41), and this effect is likely mediated by the regulation of EMT-related molecules (Slug and Twist), MMPs, and angiogenic factors (VEGFA and VEGFC; ref. 22). Our results showed that CBX4 recruits HDAC3, but not HDAC1, to the Runx2 promoter to impede cell migration and invasion in colorectal carcinoma. Furthermore, our data demonstrate that the interaction between CBX4 and HDAC3 is required for aforementioned functions in colorectal carcinoma, indicating that stabilizing the binding of CBX4 with HDAC3 with a specific agent, such as a small molecule, may be a good strategy to treat colorectal carcinoma patients harboring metastases.

No potential conflicts of interest were disclosed.

Conception and design: X. Wang, T. Kang

Development of methodology: T. Kang

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): X. Wang, L. Li, Y. Wu, G. Wang, G. Qin, T. Kang

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): X. Wang

Writing, review, and/or revision of the manuscript: X. Wang, R.-H. Xu, T. Kang

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): R. Zhang, M. Zhang, D. Liao, T. Kang

Study supervision: T. Kang

We thank both Prof. Binhua P. Zhou (University of Kentucky) and Prof. Guo-Qiang Chen (Shanghai Jiao Tong University School of Medicine, China) for the kind gifts of plasmids.

This work received support from the Yangtze River Scholarship (85000-52121100 to T. Kang), the National Key Research and Development Program of China (no: 2016YFA0500304 to T. Kang), and the National Nature Science Foundation in China (NSFC; 81530081, 31571395 to T. Kang).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Margueron
R
,
Reinberg
D
. 
The Polycomb complex PRC2 and its mark in life
.
Nature
2011
;
469
:
343
9
.
2.
Simon
JA
,
Kingston
RE
. 
Mechanisms of polycomb gene silencing: knowns and unknowns
.
Nat Rev Mol Cell Biol
2009
;
10
:
697
708
.
3.
Schoeftner
S
,
Sengupta
AK
,
Kubicek
S
,
Mechtler
K
,
Spahn
L
,
Koseki
H
, et al
Recruitment of PRC1 function at the initiation of X inactivation independent of PRC2 and silencing
.
EMBO J
2006
;
25
:
3110
22
.
4.
Gil
J
,
O'Loghlen
A
. 
PRC1 complex diversity: where is it taking us?
Trends Cell Biol
2014
;
24
:
632
41
.
5.
Liu
B
,
Liu
YF
,
Du
YR
,
Mardaryev
AN
,
Yang
W
,
Chen
H
, et al
Cbx4 regulates the proliferation of thymic epithelial cells and thymus function
.
Development
2013
;
140
:
780
8
.
6.
Luis
NM
,
Morey
L
,
Mejetta
S
,
Pascual
G
,
Janich
P
,
Kuebler
B
, et al
Regulation of human epidermal stem cell proliferation and senescence requires polycomb- dependent and -independent functions of Cbx4
.
Cell Stem Cell
2011
;
9
:
233
46
.
7.
Klauke
K
,
Radulovic
V
,
Broekhuis
M
,
Weersing
E
,
Zwart
E
,
Olthof
S
, et al
Polycomb Cbx family members mediate the balance between haematopoietic stem cell self-renewal and differentiation
.
Nat Cell Biol
2013
;
15
:
353
62
.
8.
Sauvageau
M
,
Sauvageau
G
. 
Polycomb group proteins: multi-faceted regulators of somatic stem cells and cancer
.
Cell Stem Cell
2010
;
7
:
299
313
.
9.
Li
J
,
Xu
Y
,
Long
X-D
,
Wang
W
,
Jiao
H-K
,
Mei
Z
, et al
Cbx4 governs HIF-1α to potentiate angiogenesis of hepatocellular carcinoma by its SUMO E3 ligase activity
.
Cancer Cell
2014
;
25
:
118
31
.
10.
Tang
J
,
Wang
G
,
Zhang
M
,
Li
FY
,
Sang
Y
,
Wang
B
, et al
Paradoxical role of CBX8 in proliferation and metastasis of colorectal cancer
.
Oncotarget
2014
;
5
:
10778
90
.
11.
Forzati
F
,
Federico
A
,
Pallante
P
,
Abbate
A
,
Esposito
F
,
Malapelle
U
, et al
CBX7 is a tumor suppressor in mice and humans
.
J Clin Invest
2012
;
122
:
612
23
.
12.
Zhang
XW
,
Zhang
L
,
Qin
W
,
Yao
XH
,
Zheng
LZ
,
Liu
X
, et al
Oncogenic role of the chromobox protein CBX7 in gastric cancer
.
J Exp Clin Cancer Res
2010
;
29
:
114
.
13.
Scott
CL
,
Gil
J
,
Hernando
E
,
Teruya-Feldstein
J
,
Narita
M
,
Martinez
D
, et al
Role of the chromobox protein CBX7 in lymphomagenesis
.
Proc Natl Acad Sci U S A
2007
;
104
:
5389
94
.
14.
Wotton
D
,
Merrill
JC
. 
Pc2 and SUMOylation
.
Biochem Soc Trans
2007
;
35
:
1401
4
.
15.
Kagey
MH
,
Melhuish
TA
,
Wotton
D
. 
The polycomb protein Pc2 is a SUMO E3
.
Cell
2003
;
113
:
127
37
.
16.
Fischle
W
,
Wang
Y
,
Jacobs
SA
,
Kim
Y
,
Allis
CD
,
Khorasanizadeh
S
. 
Molecular basis for the discrimination of repressive methyl-lysine marks in histone H3 by Polycomb and HP1 chromodomains
.
Genes Dev
2003
;
17
:
1870
81
.
17.
Wang
B
,
Tang
J
,
Liao
D
,
Wang
G
,
Zhang
M
,
Sang
Y
, et al
Chromobox homolog 4 is correlated with prognosis and tumor cell growth in hepatocellular carcinoma
.
Ann Surg Oncol
2013
;
20
:
S684
92
.
18.
Caldwell
S
,
Park
SH
. 
The epidemiology of hepatocellular cancer: from the perspectives of public health problem to tumor biology
.
J Gastroenterol
2009
;
44
:
96
101
.
19.
Wai
PY
,
Mi
Z
,
Gao
C
,
Guo
H
,
Marroquin
C
,
Kuo
PC
. 
Ets-1 and runx2 regulate transcription of a metastatic gene, osteopontin, in murine colorectal cancer cells
.
J Biol Chem
2006
;
281
:
18973
82
.
20.
Ye
GD
,
Sun
GB
,
Jiao
P
,
Chen
C
,
Liu
QF
,
Huang
XL
, et al
OVOL2, an inhibitor of WNT signaling, reduces invasive activities of human and mouse cancer cells and is downregulated in human colorectal tumors
.
Gastroenterology
2016
;
150
:
659
671
.
21.
Wu
Y
,
Zhou
L
,
Wang
X
,
Lu
J
,
Zhang
R
,
Liang
X
, et al
A genome-scale CRISPR-Cas9 screening method for protein stability reveals novel regulators of Cdc25A
.
Cell Discov
2016
;
2
:
16014
.
22.
Sun
SS
,
Zhang
L
,
Yang
J
,
Zhou
X
. 
Role of runt-related transcription factor 2 in signal network of tumors as an inter-mediator
.
Cancer Lett
2015
;
361
:
1
7
.
23.
Kerppola
TK
. 
Polycomb group complexes–many combinations, many functions
.
Trends Cell Biol
2009
;
19
:
692
704
.
24.
Mei
Z
,
Jiao
H
,
Wang
W
,
Li
J
,
Chen
G
,
Xu
Y
. 
Polycomb chromobox 4 enhances migration and pulmonary metastasis of hepatocellular carcinoma cell line MHCC97L
.
Sci China Life Sci
2014
;
57
:
610
7
.
25.
Yang
SH
,
Sharrocks
AD
. 
The SUMO E3 ligase activity of Pc2 is coordinated through a SUMO interaction motif
.
Mol Cell Biol
2010
;
30
:
2193
205
.
26.
Urvalek
AM
,
Gudas
LJ
. 
Retinoic acid and histone deacetylases regulate epigenetic changes in embryonic stem cells
.
J Biol Chem
2014
;
289
:
19519
30
.
27.
Li
B
,
Zhou
J
,
Liu
P
,
Hu
J
,
Jin
H
,
Shimono
Y
, et al
Polycomb protein Cbx4 promotes SUMO modification of de novo DNA methyltransferase Dnmt3a
.
Biochem J
2007
;
405
:
369
78
.
28.
Oh
Y
,
Chung
KC
. 
Small ubiquitin-like modifier (SUMO) modification of zinc finger protein 131 potentiates its negative effect on estrogen signaling
.
J Biol Chem
2012
;
287
:
17517
29
.
29.
Kang
X
,
Qi
Y
,
Zuo
Y
,
Wang
Q
,
Zou
Y
,
Schwartz
RJ
, et al
SUMO-specific protease 2 is essential for suppression of polycomb group protein-mediated gene silencing during embryonic development
.
Mol Cell
2010
;
38
:
191
201
.
30.
Tan
J
,
Jones
M
,
Koseki
H
,
Nakayama
M
,
Muntean
AG
,
Maillard
I
, et al
CBX8, a polycomb group protein, is essential for MLL-AF9-induced leukemogenesis
.
Cancer Cell
2011
;
20
:
563
75
.
31.
Long
J
,
Zuo
D
,
Park
M
. 
Pc2-mediated sumoylation of Smad-interacting protein 1 attenuates transcriptional repression of E-cadherin
.
J Biol Chem
2005
;
280
:
35477
89
.
32.
Satijn
DP
,
Olson
DJ
,
van der Vlag
J
,
Hamer
KM
,
Lambrechts
C
,
Masselink
H
, et al
Interference with the expression of a novel human polycomb protein, hPc2, results in cellular transformation and apoptosis
.
Mol Cell Biol
1997
;
17
:
6076
86
.
33.
Dahiya
A
,
Wong
S
,
Gonzalo
S
,
Gavin
M
,
Dean
DC
. 
Linking the Rb and polycomb pathways
.
Mol Cell
2001
;
8
:
557
69
.
34.
Forzati
F
,
Federico
A
,
Pallante
P
,
Fedele
M
,
Fusco
A
. 
Tumor suppressor activity of CBX7 in lung carcinogenesis
.
Cell Cycle
2012
;
11
:
1888
91
.
35.
Chimge
NO
,
Frenkel
B
. 
The RUNX family in breast cancer: relationships with estrogen signaling
.
Oncogene
2013
;
32
:
2121
30
.
36.
Chen
LF
. 
Tumor suppressor function of RUNX3 in breast cancer
.
J Cell Biochem
2012
;
113
:
1470
7
.
37.
Janes
KA
. 
RUNX1 and its understudied role in breast cancer
.
Cell Cycle
2011
;
10
:
3461
5
.
38.
Altieri
DC
,
Languino
LR
,
Lian
JB
,
Stein
JL
,
Leav
I
,
van Wijnen
AJ
, et al
Prostate cancer regulatory networks
.
J Cell Biochem
2009
;
107
:
845
52
.
39.
Endo
T
,
Ohta
K
,
Kobayashi
T
. 
Expression and function of Cbfa-1/Runx2 in thyroid papillary carcinoma cells
.
J Clin Endocrinol Metab
2008
;
93
:
2409
12
.
40.
Cohen-Solal
KA
,
Boregowda
RK
,
Lasfar
A
. 
RUNX2 and the PI3K/AKT axis reciprocal activation as a driving force for tumor progression
.
Mol Cancer
2015
;
14
:
137
.
41.
Sase
T
,
Suzuki
T
,
Miura
K
,
Shiiba
K
,
Sato
I
,
Nakamura
Y
, et al
Runt-related transcription factor 2 in human colon carcinoma: a potent prognostic factor associated with estrogen receptor
.
Int J Cancer
2012
;
131
:
2284
93
.

Supplementary data